JP6467332B2 - Adaptive equalizer and tap coefficient updating method - Google Patents

Description

  The present invention relates to an adaptive equalizer used in an OLT (Optical Line Terminal) or ONU (Optical Network Unit) provided in a PON (Passive Optical Network).

  In an adaptive equalizer, an RDE (Radius Directed Equalization) tap coefficient update algorithm has an advantage of a short convergence time with respect to a high multilevel QAM (Quadrature Amplitude Modulation) signal (for example, see Non-Patent Document 1). .)

  On the other hand, in a FIR (Finite Impulse Response) filter, the fractional interval tap configuration has the advantage that the band of the equalization transfer function can be expanded beyond the Nyquist band of the signal symbol rate. For this reason, an adaptive equalizer equipped with an FIR filter having a fractional interval tap configuration is resistant to the phase change of its operation clock (see, for example, Non-Patent Document 2).

Ready, M.M. J. et al. , Gooch, R .; P. : 'Blind equalization based on radius directed adaptation', International Conf. on Acoustics, Speech and Signal Processing (ICASSP), Albuquerque, NM, USA, April 1990, pp. 1699-1702 R. D. Gitlin and S.M. B. Weinstein, “Fractionally-spaced equalization: an implied digital transversal equalizer,” Bell Syst. Tech. J. et al. , Vol. 60, no. 2, pp. 275-296, Feb. 1981. Iiyama, N .; , Kani, J .; , Terada, J .; , And Yoshimoto, N .; : 'Feasibility study on a scheme for coexistence of DSP-based PON and 10-Gbps / λ PON using hierarchical star QAM format', J. Lightwave Technol. , 2013, 31, (18), pp. 3085-3092

  In a high-speed system of 10 Gbps or more, the equalization technique of Non-Patent Document 1 is difficult to realize because of the large amount of DSP (Digital Signal Processing) computation in tap coefficient updating.

  An object of the present invention is to reduce the amount of DSP calculation in an RDE-based adaptive equalizer.

An adaptive equalizer according to the present invention includes:
An adaptive equalizer that compensates for signal damage in an input signal modulated with multi-level QAM,
A code error calculating unit that calculates a code error consisting of a real part and an imaginary part each having only a positive or negative sign with reference to an output signal from the adaptive equalizer;
A tap coefficient calculation unit that updates a tap coefficient that compensates for signal damage of the input signal using the code error; and
Is provided.

  The code error calculation unit may derive the code error using calculation formulas (1) and (2) described later.

  The tap coefficient calculation unit may calculate the updated tap coefficient using a calculation formula (4) described later based on the code error.

The input signal has a digital value of a fractional interval in which a multi-level QAM signal is oversampled,
As a configuration for compensating for the signal damage of the input signal, it comprises a FIR filter having a fractional interval tap configuration,
The tap coefficient calculation unit may update the tap coefficient for each fractionally spaced tap using the code error.

The tap coefficient updating method according to the present invention includes:
A tap coefficient updating method executed by an adaptive equalizer that compensates for signal damage of a multi-level QAM modulated input signal,
A sign error calculation procedure for calculating a sign error consisting of a real part and an imaginary part each having only a positive or negative sign with reference to an output signal from the adaptive equalizer;
A tap coefficient calculation procedure for updating a tap coefficient that compensates for signal damage of the input signal using the code error; and
Have
In the code error calculation procedure, the code error may be derived using calculation formulas (1) and (2) described later.

  According to the present invention, since a new code error calculation algorithm that outputs in binary is used, the DSP calculation amount in the RDE-based adaptive equalizer can be reduced. As a result, the present invention can reduce the load in the equalization processing and realize signal compensation at high speed.

2 shows a configuration example of an adaptive equalizer according to the first embodiment. It is a constellation map which shows an example of an input signal. 3 shows a configuration example of a communication system according to a second embodiment. The structural example of the receiver 22 is shown.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In addition, this invention is not limited to embodiment shown below. These embodiments are merely examples, and the present invention can be implemented in various modifications and improvements based on the knowledge of those skilled in the art. In the present specification and drawings, the same reference numerals denote the same components.

  In order to reduce the DSP calculation amount of RDE, this embodiment proposes a tap coefficient update algorithm based on a signed-error scheme. The algorithm is as follows.

ただし、ｅ（ｎ）は符号エラーであり、ｅ_Ｒ（ｎ）は実数エラーであり、ｅ_Ｉ（ｎ）は虚数エラーである。 The error function of the adaptive equalizer according to the embodiment is expressed by the following equation.

However, e (n) is a sign error, e _R (n) is a real error, and e _I (n) is an imaginary error.

ただし、ｓｇｎは符号関数であり、ｙ（ｎ）は適応型等化器からの出力信号であり、Ｒ_ｋは定数であり、Ｒｅ｛ｙ（ｎ）｝は出力信号ｙ（ｎ）の実数部であり、Ｉｍ｛ｙ（ｎ）｝は出力信号ｙ（ｎ）の虚数部である。 e _R (n) and e _I (n) are represented by the following equations.

Where sgn is a sign function, y (n) is an output signal from the adaptive equalizer, R _k is a constant, and Re {y (n)} is a real part of the output signal y (n). And Im {y (n)} is the imaginary part of the output signal y (n).

Here, the constant R _k means the radius of the constellation symbol that is closest to the output signal y (n). For example, if the input signal is Star8QAM signal (FIG. 2), there are two circles with different radii _{R 1} and _{R 2.} In this case, the constant R _k is

  As represented by Equation (1) and Equation (2), the sign error e (n) is binary data having only positive or negative signs of +1 and −1. For this reason, the DSP calculation amount in the tap coefficient update can be greatly reduced. Moreover, since the error function according to the present embodiment is defined by complex numbers, it can correspond to IQ (in-phase and quadrature-phase) signals. Therefore, the adaptive equalizer according to the embodiment can also be used in a high-speed system of 10 Gbps or higher that transmits a high-multilevel QAM signal. In particular, an adaptive equalizer equipped with a FIR filter having a fractional interval tap configuration can be installed in a high-speed system of 10 Gbps or more that transmits a high-level multilevel QAM signal.

(Embodiment 1)
FIG. 1 shows a configuration example of an adaptive equalizer according to this embodiment. The adaptive equalizer 40 according to the present embodiment is an RDE adaptive equalizer to which the tap coefficient update algorithm of the present embodiment is applied. The adaptive equalizer 40 according to the present embodiment includes a tap coefficient update unit 41, an FIR filter unit 42, and a decimator 43. The tap coefficient update unit 41 has a configuration of a code error calculation unit 411 and a tap coefficient calculation unit 412.

  The input signal is oversampled by an ADC (Analog-to-Digital Converter) 32 at the front end of the receiver. For example, when the symbol period of the input signal is T, the ADC 32 samples at the sampling rate K / T and converts the analog signal into a digital signal. Thereby, the signal converted into digital becomes a fractional interval of T / K.

  A complex number converter (not shown) converts the digital signal output from the ADC 32 into a complex number digital signal. Thereby, an input vector x (n) is generated.

The code error calculation unit 411 of the tap coefficient updating unit 41 outputs a code error e (n) based on the above-described equations (1) and (2) using the constant R _k and the output signal y (n).

ここで、μはステップサイズパラメータ、＊は複素共役記号である。 The tap coefficient calculation unit 412 repeatedly calculates the obtained code error e (n) and the input vector x (n) using the following tap coefficient update expression to update the tap coefficient vector h (n).

Here, μ is a step size parameter, and * is a complex conjugate symbol.

The FIR filter unit 42 has a fractional interval tap configuration. For example, the FIR filter unit 42 includes a delay unit 421, an integrator 422, and an adder 423. Each delay unit 421 delays each digital signal included in the input vector x by a delay time T / K for each digital signal. Each integrator 422, the tap coefficients h ₀ to h _i input from the tap coefficient calculation unit 412 of the tap coefficient updating unit 41 integrates the digital signal delayed by the delay unit 421. The adder 423 adds the output values of the accumulators 422. As a result, the FIR filter unit 42 equalizes the input signal. Therefore, the adaptive equalizer 40 can compensate for signal damage of the input signal.

  Decimator 43 returns the input signal oversampled at sampling rate K / T to a digital signal at symbol rate 1 / T.

  As defined in Equations (1) and (2), the sign error e (n) is binary data having only positive or negative signs of +1 and -1. For this reason, the adaptive equalizer 40 according to the present embodiment can greatly reduce the DSP calculation amount in the tap coefficient update.

  The embodiment is not limited to the Star8QAM signal shown in FIG. 2, and can be applied to any multilevel QAM. For example, the present invention can be applied to M-ary rectangularQAM and starQAM.

(Embodiment 2)
FIG. 3 shows a configuration example of a communication system according to the present embodiment. The communication system according to the present embodiment is a PON system based on hierarchical modulation. Hierarchical modulation technology enables coexistence between a future DSP access system based on a multi-level signal and an existing OOK (On-Off Keying) system by using a hierarchically modulated QAM signal (for example, non-coding). (See Patent Document 3).

  In the communication system according to the present embodiment, the OLT 91 and a plurality of ONUs 92 are connected by PON. The OLT 91 includes a hierarchical QAM transmitter 11 and a hierarchical QAM receiver 12. The ONU 92 includes a transmitter 21 and a receiver 22. The receiver 22 of the ONU 92 # 1 receives the QAM signal. The receiver 22 of the ONU 92 # 2 receives the OOK signal.

  The algorithm according to the present embodiment can be applied to the receiver 22 of the ONU 92 # 1 in a PON system based on Star8QAM hierarchical modulation as shown in FIG. FIG. 4 shows an example of the configuration of the receiver 22. The receiver 22 includes an optical coherent receiver 31, an ADC 32, a signal processing unit 33, and a PON frame processing unit 34.

  The optical coherent receiver 31 includes an optical branching unit 311, an optical multiplexing / demultiplexing unit 312, a light receiving unit 313, a π / 2 optical delay unit 314, and an LO light source 315. The optical coherent receiver 31 outputs the I component and Q component electrical signals included in the received signal light using these. Here, the light receiving unit 313 is preferably a balanced light receiver.

  The ADC 32 converts the analog electrical signal output from the optical coherent receiver 31 into a digital signal. The I component and Q component digital signals are input to the signal processing unit 33.

  The signal processing unit 33 includes a complex number conversion unit 331, a ring ratio calculation unit 332, an equalization unit 333, and a demodulation unit 334. The equalizer 333 has the same function as the adaptive equalizer 40 described in the first embodiment.

  The complex number conversion unit 331 converts the IQ signal into a complex number complex signal Sc and outputs the complex signal Sc to the equalization unit 333. The equalization unit 333 compensates for signal damage in the complex signal Sc using the equations (1) and (2).

At that time, the ring ratio calculation unit 332 uses the complex signal Sc output from the complex number conversion unit 331 in the time interval of the first preamble of the received frame, and R1 and R2 (that is, R2) of the hierarchically modulated QAM signal. _k number) is calculated and provided to the equalization unit 333. Alternatively, the hierarchical QAM transmitter 11 of the OLT 91 may describe the R _k numerical value in the transmission frame as binary logical information. In this case, the equalization unit 333 acquires the R _k value from the PON frame processing unit 34 provided in the receiver 22 of the ONU 92 # 1.

  The present invention can be applied to the information communication industry.

11: Hierarchical QAM transmitter 12: Hierarchical QAM receiver 21: Transmitter 22: Receiver 31: Optical coherent receiver 311: Optical branching unit 312: Optical multiplexing / demultiplexing unit 313: Light receiving unit 314: π / 2 optical delay unit 315 : LO light source 32: ADC
33: signal processing unit 331: complex number conversion unit 332: ring ratio calculation unit 333: equalization unit 334: demodulation unit 34: PON frame processing unit 40: adaptive equalizer 41: tap coefficient update unit 411: code error calculation unit 412: Tap coefficient calculation unit 42: FIR filter unit 421: Delay unit 422: Accumulator 423: Adder 43: Decimator 91: OLT
92: ONU
93: Optical splitter

Claims (6)

An adaptive equalizer that compensates for signal damage of an input signal modulated by multilevel QAM (Quadrature Amplitude Modulation),
A code error calculating unit that calculates a code error consisting of a real part and an imaginary part each having only a positive or negative sign with reference to an output signal from the adaptive equalizer;
A tap coefficient calculation unit that updates a tap coefficient that compensates for signal damage of the input signal using the code error; and
An adaptive equalizer comprising: The adaptive equalizer according to claim 1, wherein the code error calculation unit derives the code error using the following calculation formula.

However, e (n) is a sign error, e _R (n) is a real number error, e _I (n) is an imaginary number error, sgn is a sign function, and y (n) is the adaptive type or the like. R _k is a constant, Re {y (n)} is the real part of the output signal y (n), and Im {y (n)} is the output signal y (n). Is the imaginary part. The tap coefficient calculation unit calculates the updated tap coefficient using the following calculation formula based on the code error:
The adaptive equalizer according to claim 1, wherein the adaptive equalizer is characterized in that

Note that h (n) is a tap coefficient vector, μ is a step size parameter, x (n) is an input vector, and * is a complex conjugate symbol. The input signal has a digital value of a fractional interval in which a multi-level QAM signal is oversampled,
As a configuration for compensating for signal damage of the input signal, an FIR (Finite Impulse Response) filter having a fractional interval tap configuration is provided,
The tap coefficient calculation unit updates the tap coefficient for each fractionally spaced tap using the code error.
The adaptive equalizer according to any one of claims 1 to 3. A tap coefficient updating method executed by an adaptive equalizer that compensates for signal damage of an input signal modulated by multi-level QAM (Quadrature Amplitude Modulation),
A sign error calculation procedure for calculating a sign error consisting of a real part and an imaginary part each having only a positive or negative sign with reference to an output signal from the adaptive equalizer;
A tap coefficient calculation procedure for updating a tap coefficient that compensates for signal damage of the input signal using the code error; and
A tap coefficient updating method. 6. The tap coefficient updating method according to claim 5, wherein in the code error calculation procedure, the code error is derived using the following calculation formula.

Where e (n) is a sign error and e _R (N) is a real number error and e _I (N) is an imaginary error, sgn is a sign function, y (n) is an output signal from the adaptive equalizer, R _k Is a constant, Re {y (n)} is the real part of the output signal y (n), and Im {y (n)} is the imaginary part of the output signal y (n).

Images